Incorporating built environment features in Air Passenger Demand forecasting: A spatial econometric approach for enhancing transport policy and planning

This study explores the determinants of Air Passenger Demand (APD) in India, with a special focus on the role of built environment features alongside geoeconomic and service-related variables. For spatially distributed data of APD, the study attempts to incorporate the spatial interaction effects in the APD model development using spatial econometric techniques with the Bayesian Markov Chain Monte Carlo approach. First, the spatial patterns and intensities of APD are visualized using a desire line diagram, and the cities are classified as low, medium and high to understand the rationale behind APD. Then, Moran's I statistics is used to confirm the presence of spatial autocorrelation of determinants with clustering patterns, and significant variables are used as spatial indicators in the development of the APD model. Also, Local Indicator of Spatial Autocorrelation (LISA) analysis is utilised to understand the dynamic localised spatial information. APD models are developed, and how individual, interaction and combined effects influence APD are explained. The Spatial Durbin Model (SDM) outperformed with a 0.622 posterior model probability for combined interaction effects (geoeconomic, service-related and built environment). The model's significant spatial autoregressive coefficient (α = 0.358) confirms strong spatial interdependence, and the study uses impact decomposition to separate the determinants' direct (own-city) and indirect (spillover) effects. Strategic policies based on this decomposition are proposed to capitalize on these study insights and drive sustainable APD growth in the aviation sector.